1. Field of the Invention
The present invention relates to a method for amplifying optical signals in a multi-stage Raman amplifier and to a multi-stage Raman amplifier. In particular, the present invention relates to a method for amplifying optical signals in a lumped multi-stage Raman amplifier and to a lumped multi-stage Raman amplifier.
2. Description of the Related Art
The maximum number of dense wavelength-division-multiplexed (DWDM) signals that can be transmitted over a single optical fiber has been rapidly increasing over the last few years. This trend, coupled with an increasing data rate per signal, has lead to a profound increase in the amount of signal power propagating through such optical fibers, in order to sustain applications such as data communications and the Internet. This has created a simultaneous demand for large bandwidth and high output power from the optical amplifiers used in such systems.
Erbium-doped fiber amplifiers (EDFAs) are a relatively mature technology. The amount of bandwidth that such amplifiers can produce, however, is fundamentally limited by the physics of the erbium atoms that produce the optical gain in such devices.
Raman amplifiers offer an alternative to EDFAs and are recently attracting much attention in DWDM systems, due to their distinctive flexibility in bandwidth design and growing maturity of high-power pump module technology. Raman amplifiers offer several advantages: low noise, flexible use of signal wavelengths (since the Raman gain peak is mainly dependent on the pump wavelength and not on the emission cross section of a dopant) and a broad gain bandwidth (multiple pumps can be employed). In particular, multi-wavelength pumping allows to extend the wavelength range over which flat Raman gain can be achieved: the total gain profile of such amplifiers consists of a superposition of the contributions from each individual pump.
On the other hand, many factors must be considered in the design of the amplifier and systems that use them. A thorough understanding of some key factors is required, such as, for example, pump-to-pump power transfer, signal-to-signal power transfer, pump depletion, double Rayleigh scattering (DRS) and amplifier spontaneous noise. H. Kidorf et al., in their article “Pump interactions in a 100-nm Bandwidth Raman amplifier”, IEEE Photonics Technology Letters, vol. 11, no. 5, pag. 530-2 (1999), disclose a computer model that simulates all the physical properties that affect the above listed factors. The computer model numerically solves a differential equation. The authors used their model in order to design a distributed Raman amplifier with a 100-nm bandwidth and with minimum gain variation. The amplifier was designed to have a total output power of 50 mW for 100 channels spaced 1 nm/channel. The intended use of the amplifier was the compensation of 45-km fiber spans (intended for 10,000 km transmission) made of pure silica core fiber plus an extra 3 dB to compensate for internal losses (WDM coupler, isolator, etc.). In a first attempt, the authors tried to evenly space eight pumps between 1432 and 1516 nm. With the goal of implementing the amplifier with semiconductor pumps, a maximum pump power of 120-130 mW per pump was chosen. According to the authors, the result of this simplistic design was a very poor amplifier: the gain variation of the amplifier was 10.5 dB due to power being transferred from the low wavelength pumps to the high wavelengths pumps. The added power in the high wavelengths pumps caused excessive gain at the higher signal wavelengths. Through iterative modeling, the authors arrived at a pump scheme whereby a large energy density at low wavelengths provides pump power for both the high wavelength pumps and the low wavelength signals. By properly balancing the pump's spectral density, an amplifier with a peak-to-peak gain ripple of 1.1 dB was designed.
Another known approach for obtaining a flat Raman gain on broad bandwidths using multi-wavelength pumping is to carefully select the magnitude of each contribution in order to achieve the desired gain profile. For example, P. M. Krummrich et al, “Bandwidth limitations of broadband distributed Raman fiber amplifiers for WDM systems”, OFC 2001, vol. 1 pag. MI3/1-3, analyze the impact of pump interactions both numerically and experimentally. Their numerical model works by integrating a set of coupled differential equations describing the propagation of pump and signal radiation in the transmission fiber. For the experiments, the authors use a multi-channel pump unit. The pump radiation in the wavelength range of 1409-1513 nm is generated by high power laser diodes and combined by a WDM coupler. More particularly, seven pump channels have been used in a counter-directional pump configuration to achieve flat gain in the wavelength range of 1530-1605 nm with the following set of pump wavelengths: 1424, 1438, 1453, 1467, 1483, 1497 and 1513 nm. The launch powers have been adjusted to achieve an average gain of 10 dB with a gain variation of less than 0.5 dB. According to the authors, the strongest impact of pump interactions can be observed for the channels with the shortest and longest wavelength. The pump channel at 1424 nm experiences 11 dB of additional loss and the channel at 1513 nm experiences 7 dB gain. Further, according to the authors, should the overall gain be increased, it is quite difficult to predict which pump diode output power has to be increased by which amount, due to the energy transfer between the pumps. For gain values higher than 10 dB, due to the strong impact of pump interactions and the resulting gain tilt, the gain at the long wavelength side always grows stronger than the gain at the short wavelength side if any of the pump laser output powers is increased. According to the authors, this effect limits the maximum value of flat gain for a signal wavelength range of 75 nm to approximately 19 dB.
Another numerical model is disclosed in X. Zhou et al., “A Simplified Model and Optimal Design of a Multiwavelength Backward-Pumped Fiber Raman Amplifier”, IEEE Photonics Technology Letters, vol. 13, no. 9, pag. 945-7 (2001). The authors obtain a closed-form analytical expression for pump power evolution. Based on the obtained analytical expression, formulas for calculating the small-signal optical gain and noise figure are then presented. The application of the developed model in pump optimization design is also discussed. In order to obtain the design for a wide-band optical amplifier having a flat gain, the authors use the following parameters: maximal pump light frequency 214.2 THz (1400 nm), minimal pump light frequency 200 THz (1500 nm), fiber loss at the pump frequency 0.3 dB/km, fiber loss at the signal frequency 0.2 dB/km, fiber length 10 km, required gain 20 dB, number of channels 100 (from 1510 to 1610 with 1 nm separation), fiber effective area 50 μm2. By considering three pump wavelengths, the authors obtain the optimal pump wavelength at 1423, 1454 and 1484 nm, and the corresponding optimal pump power as 1.35, 0.19 and 0.20 W, respectively. By considering six pump wavelengths, the authors obtain the optimal pump wavelength as 1404, 1413, 1432, 1449, 1463 and 1495 nm, and the corresponding optimal pump power as 0.68, 0.6, 0.44, 0.19, 0.076 and 0.054 W, respectively. It is shown that the gain ripple can be compressed definitely by increasing the number of pump light sources. However, the noise performance of the six-pump amplifier is worse than that of the three-pump amplifier. According to the authors, this is because the six-pump amplifier has a pump at higher wavelength (1495 nm). More particularly, FIG. 1 of the article shows a noise figure ranging from about 7.5 dB at 1510 nm to about 4 dB at 1610 nm for the six-pump amplifier. On the other hand, FIG. 1 of the article shows a gain variation of about 5 dB in the whole wavelength range for the three-pump amplifier.
U.S. patent application Ser. No. 2002/0044335 discloses an amplifier apparatus including an optical transmission line with a Raman amplification region that provides a pump to signal power conversion efficiency of at least 20%. The Raman amplification region is configured to amplify a signal with multiple wavelengths over at least a 30 nm range of wavelengths, preferably over at least a 50 nm range of wavelengths, more preferably over at least a 70 nm range of wavelengths. A pump source is coupled to the optical transmission line. An input optical signal is amplified in the Raman amplification region and an output signal is generated that has at least 100 mW more power than the input optical signal. In one disclosed embodiment, the amplifier apparatus had more than 3.2 dB of gain over 105 nm utilizing a Lucent DK-20 dispersion compensating fiber. 1 dB of loss was assumed to be present at both ends of the gain fiber. This fiber was pumped with 250 mW at 1396, 1416 and 1427 nm, 150 mW at 1450 nm, 95 mW at 1472 nm and 75 mW at 1505 nm. Ten input signals at 1520, 1530, 1540, 1550, 1560, 1570, 1580, 1590, 1600 and 1610 nm had 12 mW of power and were counter-propagating with respect to the pump wavelengths. FIG. 6 of the '44335 patent application shows the noise figure of this amplifier, ranging from about 7.5 dB at 1520 nm to about 6 dB at 1610 nm. In an embodiment disclosed as “all band Raman amplifier”, the amplifier apparatus includes a transmission line with two Raman amplification regions. Two WDMs are provided. Shorter signal wavelengths can receive more gain in one of Raman amplification regions than in the other. A lossy member can be positioned between the Raman amplification regions. The lossy member can include at least one of an add/drop multiplexer, a gain equalization element, an optical isolator or a dispersion compensating element. One WDM receives a first set of pump wavelengths provided by a first pump source and the WDM between the two amplification regions receives a second set of pump wavelengths which can provide gain to the optical signal and extract optical energy from at least a portion of the first set of pump wavelengths. The second set of wavelengths is provided by a second pump source. The WDM between the two amplification regions can substantially pass signal wavelengths as well as at least a portion of the first set of pump wavelengths between the two Raman amplification regions. According to the authors, a gain flatness of the all band Raman amplifier can be optimized by a gain flattening filter, and/or by pump wavelengths, pump powers, the number of pumps and the lengths of Raman gain fibers.
JP patent application No. 2001-109026 discloses a fiber Raman light amplifier using tellurite glass as a light amplification medium. More particularly, it discloses a fiber Raman amplifier constituted by using three lines of tellurite fibers whose lengths are respectively 150 m and three exciting light sources whose wavelengths are respectively 1370 nm, 1400 nm and 1430 nm, and by connecting three units in which the tellurite glass fibers and the exciting light sources are respectively combined in series and the amplifier is excited by making respective input power of exciting lights to be 200 mW (totally 600 mW). A gain equal to or larger than 23 dB is obtained in a 100 nm band extending from 1.5 to 1.6 μm.
A further problem related to Raman amplifiers is that backward and multi-wavelength pumping scheme results in larger noise figures for shorter wavelength signals, due to pump-to-pump stimulated Raman scattering, in addition to thermal noise and wavelength-dependence of fiber attenuation coefficients. C. R. S. Fludger et al., “Fundamental Noise Limits in Broadband Raman Amplifiers”, OFC 2001, MA5/1-3, show that broadband discrete Raman amplifiers based on silica-germania will have a noise figure significantly greater than the quantum limit. According to the authors, in a discrete Raman amplifier the wavelength dependence of the noise figure is determined by four main factors. These include the gain spectrum and component losses at the amplifier input. Also, stimulated Raman scattering (SRS) will transfer power from the shorter wavelength pumps to the longer wavelength pumps. Finally, noise figure is also affected by increased spontaneous emission due to the thermal distribution of phonons in the ground state. In particular, if a pump provides a large amount of gain to a closely spaced signal there will be a large increase in the excess spontaneous noise. If the ratio of the gain given by that pump to the overall total gain from all the pumps is small, the noise figure will tend to 3 dB. The authors evaluate the best achievable internal noise figure for a five wavelength-pumped discrete Raman amplifier. The relative gains from each pump were chosen to give the broadest and flattest spectrum with the highest pump wavelength at 1495 nm and the lowest signal near 1500 nm. It is shown that since a substantial amount of the total gain at shorter signal wavelengths is given by the 1495 nm pump, there is increased spontaneous emission as the signals approach the pump. At room temperature, the internal noise figure of the amplifier is between 5 and 6 dB below 1520. However, the total noise figure of the amplifier will be greater than this once fiber loss and the insertion loss of components at the amplifier input are included.
S. Kado et al., “Broadband flat-noise Raman amplifier using low-noise bi-directionally pumping sources”, ECOC 2001, propose and experimentally demonstrate an optimized bi-directional pumping scheme that realizes a less than 0.7 dB flatness over C- and L-bands of both Raman gain and optical noise figure, simultaneously. In order to use forward-pumping, for the proposed method, a new type of pump laser having low relative intensity noise (RIN) is also developed. Such laser is a wavelength-stabilized multimode pump laser, where the laser chip has an internal grating layer along laser cavity for selecting more than three longitudinal modes. The RIN of the developed laser is more than 20 dB lower than a usual fiber Bragg grating stabilized laser. According to the authors, this development allows to use forward pumping scheme without significantly hurting the signal quality due to poor RIN characteristics.
Known configurations of Raman amplifiers, such as those presented above, may achieve high and flat gain in broad wavelength ranges. However, typically this goes with an unbalanced noise figure, having higher values for shorter signal wavelengths and lower values for longer signal wavelengths (except for the amplifier disclosed by Kado et al., that use a special type of pump laser in order to obtain a flat noise figure). The Applicant notices that the higher noise figure of shorter signal wavelengths may raise problems in some configurations of optical system including Raman amplifiers, in that at least a portion of the shorter signal wavelengths may risk going outside system specifications.
Further, known configurations of Raman amplifiers, especially those achieving a flat gain on a broad band of wavelengths, typically use very high power values for shorter pump wavelengths (in excess of 500 mW), due to transfer of energy between shorter pump wavelengths and longer pump wavelengths. The Applicant notices that this is not an optimal solution. In fact, reliable semiconductors lasers having a power emission in excess of 500 mW are now hardly available on the market and/or costly, so that multiple lasers having lower power emission should be used. As a consequence, the cost of the overall amplifier and/or the space occupied by the pump sources may disadvantageously increase.